在选择其他列时使用基于函数的索引

Question

我有一个按预期工作的基于 Oracle 18c 函数的索引：

1.Create 一个接受 user-defined type 对象和 returns 坐标列表作为文本的自定义函数：

create or replace function endpoint_list(shape in sde.st_geometry) return varchar2 
deterministic is
    coord_list varchar2(4000);
begin
    coord_list := 
      --sde.st_geometry functions are notoriously slow.  
      sde.st_geometry_operators.st_x_f(sde.st_geometry_operators.st_startpoint_f(sde.st_geometry_operators.st_geometryn_f(shape,1))) || ',' || 
            sde.st_geometry_operators.st_y_f(sde.st_geometry_operators.st_startpoint_f(sde.st_geometry_operators.st_geometryn_f(shape,1))) || ',' || 
                sde.st_geometry_operators.st_x_f(sde.st_geometry_operators.st_endpoint_f(  sde.st_geometry_operators.st_geometryn_f(shape,1))) || ',' ||
                    sde.st_geometry_operators.st_y_f(sde.st_geometry_operators.st_endpoint_f(  sde.st_geometry_operators.st_geometryn_f(shape,1)));
return coord_list; 
end;

2.Create 自定义函数上的基于函数的索引 (FBI)：

create index atn_endpoint_list_idx on my_owner.active_transportation(my_owner.endpoint_list(shape));

3.Run在SELECT子句中使用FBI的查询：

select  
    endpoint_list(shape) as list
from
    active_transportation --15,000 rows
where
    endpoint_list(shape) is not null


----------------------------------------------------------------------------------------------
| Id  | Operation            | Name                  | Rows  | Bytes | Cost (%CPU)| Time     |
----------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT     |                       |   727 |   299K|    50  (10)| 00:00:01 |
|*  1 |  INDEX FAST FULL SCAN| ATN_ENDPOINT_LIST_IDX |   727 |   299K|    50  (10)| 00:00:01 |
----------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
   1 - filter("my_owner"."ENDPOINT_LIST"("SHAPE") IS NOT NULL)

FBI 按预期工作。当我 select endpoint_list(shape) 作为列时，查询使用索引，将运行时间从 65 秒减少到 .09 秒。好

除了 selecting endpoint_list(shape)，我还想 select 来自 table 的其他列，例如 OBJECTID 列：

select  
    objectid,  --Note: OBJECTID has a index of it's own (unique).
    endpoint_list(shape) as list
from
    active_transportation
where
    endpoint_list(shape) is not null

-------------------------------------------------------------------------------------------
| Id  | Operation         | Name                  | Rows  | Bytes | Cost (%CPU)| Time     |
-------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |                       |   727 |   299K|   181   (4)| 00:00:01 |
|*  1 |  TABLE ACCESS FULL| ACTIVE_TRANSPORTATION |   727 |   299K|   181   (4)| 00:00:01 |
-------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
   1 - filter("my_owner"."ENDPOINT_LIST"("SHAPE") IS NOT NULL)

现在，没有使用联邦调查局。查询改为进行完整的 table 扫描，这不是我想要的。

问题：

如何在 select 添加其他列时利用基于函数的索引？

Answer 1

解决方案似乎是将附加列 (OBJECTID) 添加到索引中 — 作为复合索引：

create index atn_endpoint_list_idx on
    my_owner.active_transportation(my_owner.endpoint_list(shape),OBJECTID);
--

现在，正在使用 FBI：

----------------------------------------------------------------------------------------------
| Id  | Operation            | Name                  | Rows  | Bytes | Cost (%CPU)| Time     |
----------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT     |                       |   727 |   299K|    53  (10)| 00:00:01 |
|*  1 |  INDEX FAST FULL SCAN| ATN_ENDPOINT_LIST_IDX |   727 |   299K|    53  (10)| 00:00:01 |
----------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
   1 - filter("my_owner"."ENDPOINT_LIST"("SHAPE") IS NOT NULL)

在选择其他列时使用基于函数的索引

Utilize a function-based index while selecting additional columns

sql

oracle

indexing

query-optimization

oracle18c